ocawari 0.9.7

data/lib/ocawari/strategy/girls_news.rb

@@ -0,0 +1,19 @@
+module Ocawari
+  module Strategy
+    class GirlsNews < Parser
+
+      private
+
+      CSS_SELECTORS = [
+        "div#img_field div.single_img_field_l a",
+        "div#img_field ul li.single_img_field_s a"
+      ]
+
+      def parse
+        page.css(CSS_SELECTORS.join(", ")).map do |a|
+          File.join("http://www.stereosound.co.jp", a["href"])
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/google_plus.rb

@@ -0,0 +1,30 @@
+module Ocawari
+  module Strategy
+    class GooglePlus < Parser
+      def initialize(uri)
+        if USER_IDENTIFIER.match?(uri.to_s)
+          @uri = Addressable::URI.parse(uri.to_s.sub(USER_IDENTIFIER, ""))
+        else
+          @uri = uri
+        end
+
+        @page = Nokogiri::HTML(open(@uri).read)
+      rescue OpenURI::HTTPError
+        @page = nil
+      end
+
+      private
+
+      USER_IDENTIFIER = /u\/\d+\//
+
+      def parse
+        album_url = File.join("https://plus.google.com", @page.to_html[/(\/photos\/\d+\/albums\/\d+)/, 1])
+        album_page = Nokogiri::HTML(open(album_url).read)
+
+        images = album_page.css("img").map do |img|
+          img["src"].sub(/=w\d+-h\d+/, "=s0")
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/hustlepress.rb

@@ -0,0 +1,18 @@
+module Ocawari
+  module Strategy
+    class Hustlepress < Parser
+
+      private
+
+      CSS_SELECTORS = [
+        "div.post_content img.size-full"
+      ]
+
+      def parse
+        page.css(CSS_SELECTORS.join(", ")).map do |img|
+          img["src"]
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/imgur.rb

@@ -0,0 +1,14 @@
+module Ocawari
+  module Strategy
+    class Imgur < Parser
+
+      private
+
+      def parse
+        image_links = page.css("div.post-images a.zoom")
+
+        image_links.map { |img| "http:#{img["href"]}" }
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/instagram.rb

@@ -0,0 +1,37 @@
+module Ocawari
+  module Strategy
+    class Instagram < Parser
+      def initialize(uri)
+        if /\?taken-by=/.match?(uri.path)
+          uri.path = uri.path.sub(/\/\?taken-by=.*/, "")
+          @page = Nokogiri::HTML(open(uri).read)
+        else
+          @page = Nokogiri::HTML(open(uri).read)
+        end
+      rescue OpenURI::HTTPError
+        @page = nil
+      end
+      
+      private
+
+      def parse
+        script_tag = page.css("script").find { |script| script.text.include?("window._sharedData") }
+
+        graphql_state = script_tag.text.
+          sub("window._sharedData = ", "").
+          sub(/;$/, "").
+          yield_self { |raw| JSON.parse(raw) }
+
+        root = graphql_state.dig( "entry_data", "PostPage", 0, "graphql", "shortcode_media")
+
+        if graph_images_nodes = root.dig("edge_sidecar_to_children", "edges")
+          graph_images_nodes.map do |graph_image| 
+            graph_image.dig("node", "display_url")
+          end
+        else
+          [root["display_resources"].last["src"]]
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/kaiyou.rb

@@ -0,0 +1,25 @@
+module Ocawari
+  module Strategy
+    class Kaiyou < Parser
+
+      private
+
+      def parse
+        all_images = [header_image] + content_images.to_a
+        all_images.map do |img|
+          uri = Addressable::URI.parse(img["src"])
+
+          "#{uri.scheme}://#{uri.hostname}/press/img/#{uri.path.split("/").last}"
+        end
+      end
+
+      def header_image
+        page.at_css("div.m-article-eyecatch img")
+      end
+
+      def content_images
+        page.css("div.m-article-main img.size-full")
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/keyakizaka46.rb

@@ -0,0 +1,23 @@
+module Ocawari
+  module Strategy
+    class Keyakizaka46 < Parser
+      def initialize(uri)
+        @uri = uri
+        @page = Nokogiri::HTML(
+          open(uri, {"User-Agent" => Ocawari::WINDOWS_CHROME_USER_AGENT}).read
+        )
+
+      rescue OpenURI::HTTPError
+        @page = nil
+      end
+
+      private
+
+      def parse
+        page.css("div.box-article img").map do |img|
+          File.join("http://www.keyakizaka46.com", img["src"])
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/line.rb

@@ -0,0 +1,31 @@
+module Ocawari
+  module Strategy
+    class Line < Parser
+
+      private
+
+      CSS_SELECTOR_HIERARCHY = %w(
+        div.article-body
+        div.article-body-inner
+        img.pict
+      ).join(" ")
+
+      def parse
+        image_nodes = page.css(CSS_SELECTOR_HIERARCHY)
+
+        image_nodes.map do |img|
+          imgname = img["src"]
+
+          case imgname
+          when /-s\./
+            imgname.sub("-s.", ".")
+          when /\/small$/
+            imgname.sub("/small", "")
+          else
+            imgname
+          end
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/mantan_web.rb

@@ -0,0 +1,27 @@
+module Ocawari
+  module Strategy
+    class MantanWeb < Parser
+
+      private
+
+      CSS_SELECTORS = [
+        "ul.newsbody__thumblist li.newsbody__thumb img"
+      ]
+
+      def parse
+        amount_of_images = page.at_css("span.newsbody__photo-num").text.to_i
+        main_image = page.at_css("div.newsbody__img img")["src"]
+
+        (1..amount_of_images).to_a.map do |i|
+          if i < 10
+            main_image.sub("001_size6", "00#{i}_size10")
+          elsif i >= 10 && i < 100
+            main_image.sub("001_size6", "0#{i}_size10")
+          elsif i >= 100 && i < 1000
+            main_image.sub("001_size6", "#{i}_size10")
+          end
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/mens_fashion.rb

@@ -0,0 +1,20 @@
+module Ocawari
+  module Strategy
+    class MensFashion < Parser
+
+      private
+
+      CSS_SELECTORS = [
+        "p img.size-full",
+        "img.image_ll",
+        "img.image_ls"
+      ]
+
+      def parse
+        page.css(CSS_SELECTORS.join(", ")).map do |img|
+          img["src"]
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/modelpress.rb

@@ -0,0 +1,25 @@
+module Ocawari
+  module Strategy
+    class ModelPress < Parser
+
+      private
+
+      CSS_SELECTORS = [
+        "div#body-top img.outputthumb",
+        "article.mdpr-article img.outputthumb"
+      ]
+
+      def parse
+        image_links = page.css(CSS_SELECTORS.join(", ")).map { |img| img["src"] }
+
+        image_links.map do |link|
+          link, _query_params = link.split("?")
+
+          # width is 6000 to make Fastly return
+          # the largest image possible
+          "#{link}?width=6000&quality=100"
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/nana_bun_no_nijuuni.rb

@@ -0,0 +1,18 @@
+module Ocawari
+  module Strategy
+    class NanaBunNoNijuuni < Parser
+
+      private
+
+      CSS_SELECTORS = [
+        "div.blog_main img"
+      ]
+
+      def parse
+        page.css(CSS_SELECTORS.join(",")).map do |img|
+          File.join("http://blog.nanabunnonijyuuni.com", img["src"])
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/nana_go_go.rb

@@ -0,0 +1,18 @@
+module Ocawari
+  module Strategy
+    class NanaGoGo < Parser
+
+      private
+
+      def parse
+        img = page.css("img").find { |img| img["alt"] == "投稿画像" }
+
+        if target_img = page.at("img[alt='投稿画像']")
+          [target_img["data-src"]]
+        else
+          []
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/natalie.rb

@@ -0,0 +1,18 @@
+module Ocawari
+  module Strategy
+    class Natalie < Parser
+
+      private
+
+      CSS_SELECTORS = [
+        "div.NA_articleUnit ul.NA_imageList span.NA_thumb"
+      ]
+
+      def parse
+        page.css(CSS_SELECTORS.join(", ")).map do |image|
+          image["data-bg"].sub(/fit_\d+x\d+/, "fixw_730_hq")
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/news_dwango.rb

@@ -0,0 +1,18 @@
+module Ocawari
+  module Strategy
+    class NewsDwango < Parser
+
+      private
+
+      CSS_SELECTORS = [
+        "#js-fancy-slider-sub img"
+      ]
+
+      def parse
+        page.css(CSS_SELECTORS.join(",")).map do |img|
+          img["src"].sub("/lg_", "/")
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/nikkan_sports.rb

@@ -0,0 +1,22 @@
+module Ocawari
+  module Strategy
+    class NikkanSports < Parser
+
+      private
+
+      CSS_SELECTORS = [
+        "article#columnMain div.column-photo-area img"
+      ]
+
+      def parse
+        page.css(CSS_SELECTORS.join(", ")).map do |image|
+          path = image["style"].
+                   sub("background-image: url(", "").
+                   sub(/\);.*$/, "")
+
+          File.join("www.nikkansports.com", path)
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/no_match.rb

@@ -0,0 +1,12 @@
+module Ocawari
+  module Strategy
+    class NoMatch
+      def initialize(uri)
+      end
+      
+      def execute
+        []
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/okmusicjp.rb

@@ -0,0 +1,19 @@
+module Ocawari
+  module Strategy
+    class OkMusicJP < Parser
+
+      private
+
+      CSS_SELECTORS = [
+        "div#page_frame div.center-page_box div.head-img_box img",
+        "div#page_frame div.center-page_box img.main-text_image"
+      ]
+
+      def parse
+        page.css(CSS_SELECTORS.join(", ")).map do |img|
+          img["src"]
+        end
+      end
+    end
+  end
+end

data/lib/ocawari/strategy/sirabee.rb

@@ -0,0 +1,29 @@
+module Ocawari
+  module Strategy
+    class Sirabee < Parser
+
+      private
+
+      LARGEST_KNOWN_WIDTH = 768
+      LARGEST_KNOWN_HEIGHT = 512
+
+      def parse
+        if has_gallery_images?
+          page.css("div.entryGallery div.entryGallery-item img").map do |img|
+            img["src"].sub(/-\d{3}x\d{3}/, "-#{LARGEST_KNOWN_WIDTH}x#{LARGEST_KNOWN_HEIGHT}")
+          end
+        else
+          page.css("article.entryContent section.entryContentBody img.size-large").map do |img|
+            img["src"]
+          end
+        end
+      end
+
+      private
+
+      def has_gallery_images?
+        page.css("div.entryGallery div.entryGallery-item img").any?
+      end
+    end
+  end
+end